test/rbbt/sources/test_biomart.rb in rbbt-sources-3.3.0 vs test/rbbt/sources/test_biomart.rb in rbbt-sources-3.4.1

@@ -1,44 +1,82 @@
 require File.expand_path(File.dirname(__FILE__) + '/../../test_helper')
 
 require 'rbbt/sources/pubmed'
 require 'test/unit'
 require 'rbbt/sources/biomart'
+require 'rbbt/sources/organism'
 require 'rbbt/util/tmpfile'
 require 'test/unit'
 
 class TestBioMart < Test::Unit::TestCase
 
   def setup
-    BioMart.set_archive Organism.default_code("Hsa")
+    BioMart.set_archive "feb2014"
   end
 
   def teardown
     BioMart.unset_archive
   end
 
-  def _test_get
+  def test_get_Sce
     assert_raise BioMart::QueryError do 
       BioMart.get('scerevisiae_gene_ensembl','entrezgene', ['protein_id'],['with_unknownattr'])
     end
 
-    data = BioMart.get('scerevisiae_gene_ensembl','entrezgene', ['protein_id'],[], nil, :nocache => false, :merge => true, :wget_options => {:quiet => false})
+    BioMart.set_archive "feb2023-fungi"
+    data = BioMart.get('scerevisiae_eg_gene','entrezgene_id', ['protein_id'],[], nil, :nocache => true, :merge => true, :wget_options => {:quiet => false})
     tsv = TSV.open data, :double, :merge => true
-    assert(tsv['852236'][0].include? 'CAA84864')
+    assert(tsv['852236'][0].include? 'CAA84864.1')
 
-    data = BioMart.get('scerevisiae_gene_ensembl','entrezgene', ['external_gene_id'],[], data, :nocache => false, :wget_options => { :quiet => false} )
+    data = BioMart.get('scerevisiae_eg_gene','entrezgene_id', ['external_gene_id'],[], data, :nocache => false, :wget_options => { :quiet => false} )
     tsv = TSV.open data, :double, :merge => true
     assert(tsv['852236'][1].include? 'YBL044W')
   end
 
-  def _test_query
+  def test_get_Hsa
+    Log.severity = 0
+    data = BioMart.get('hsapiens_gene_ensembl','entrezgene', ['protein_id'],[], nil, :nocache => true, :merge => true, :wget_options => {:quiet => false})
+    tsv = TSV.open data, :double, :merge => true
+    assert(tsv['852236'][0].include? 'CAA84864.1')
+  end
+
+
+  def test_query
     data = BioMart.query('scerevisiae_gene_ensembl','entrezgene', ['protein_id','refseq_peptide','external_gene_id','ensembl_gene_id'], [], nil, :nocache => false, :wget_options => { :quiet => false})
     assert(data['852236']['external_gene_id'].include? 'YBL044W')
 
     TmpFile.with_file do |f|
       filename = BioMart.query('scerevisiae_gene_ensembl','entrezgene', ['protein_id','refseq_peptide','external_gene_id','ensembl_gene_id'], [], nil, :nocache => false, :wget_options => { :quiet => false}, :filename => f)
       data = TSV.open Open.open(filename)
       assert(data['852236']['external_gene_id'].include? 'YBL044W')
+    end
+  end
+
+  def __test_transcrip_exons
+    Log.with_severity 1 do
+      TmpFile.with_file do |f|
+        fields = ['ensembl_transcript_id','ensembl_exon_id','rank']
+        main = fields[0]
+        attrs = fields.values_at(1, 2)
+        attrs_first = [attrs.first]
+        attrs_last = [attrs.last]
+        database = 'hsapiens_gene_ensembl'
+
+        filename = BioMart.get(database, main, attrs, {"ensembl_transcript_id" => ['ENST00000357654']}, nil, :nocache => false, :wget_options => {:quiet => false}, :filename => f)
+        ppp Open.read(filename)
+
+        filename = BioMart.get(database, main, attrs_first, {"ensembl_transcript_id" => ['ENST00000357654']}, nil, :nocache => false, :wget_options => {:quiet => false}, :filename => f)
+        ppp Open.read(filename)
+
+        filename = BioMart.get(database, main, attrs_last, {"ensembl_transcript_id" => ['ENST00000357654']}, nil, :nocache => false, :wget_options => {:quiet => false}, :filename => f)
+        ppp Open.read(filename)
+
+        filename = BioMart.query(database, main, attrs, {"ensembl_transcript_id" => ['ENST00000357654']}, nil, :nocache => true, :wget_options => {:quiet => false}, :filename => f)
+        ppp Open.read(filename)
+
+        data = TSV.open Open.open(filename)
+        assert(data['852236']['external_gene_id'].include? 'YBL044W')
+      end
     end
   end
 
   def test_tsv
     data = BioMart.tsv('scerevisiae_gene_ensembl',['Entrez Gene', 'entrezgene'], [['Protein ID', 'protein_id'],['RefSeq Peptide','refseq_peptide']], [], nil, :nocache => false, :wget_options => { :quiet => false})